ggml-webgpu: fix buffer aliasing for ssm_scan and refactor aliasing logic#22456
Merged
Conversation
Contributor
|
Thanks @reeselevine! I just tested it on my Intel Arc B580 and Apple M2, and everything looks good on my end. |
ggerganov
approved these changes
Apr 28, 2026
CISC
approved these changes
Apr 28, 2026
Member
|
@SharmaRithik Nice catch on the bug! @reeselevine The approach makes sense, and the fix for rms_norm_mul is nice. Thank you both! |
IntelNav
pushed a commit
to IntelNav/llama.cpp
that referenced
this pull request
Apr 29, 2026
…ogic (ggml-org#22456) * Refactor buffer aliasing to be part of shader lib decisions * cleanup * formatting
IntelNav
pushed a commit
to IntelNav/llama.cpp
that referenced
this pull request
Apr 29, 2026
…ogic (ggml-org#22456) * Refactor buffer aliasing to be part of shader lib decisions * cleanup * formatting
cnsiva
pushed a commit
to saas-home/llama.cpp
that referenced
this pull request
Apr 29, 2026
…ogic (ggml-org#22456) * Refactor buffer aliasing to be part of shader lib decisions * cleanup * formatting
samuraieng
pushed a commit
to samuraieng/llama.cpp
that referenced
this pull request
May 6, 2026
…ogic (ggml-org#22456) * Refactor buffer aliasing to be part of shader lib decisions * cleanup * formatting
ljubomirj
pushed a commit
to ljubomirj/llama.cpp
that referenced
this pull request
May 6, 2026
…ogic (ggml-org#22456) * Refactor buffer aliasing to be part of shader lib decisions * cleanup * formatting
meh
pushed a commit
to meh/llama.cpp
that referenced
this pull request
May 10, 2026
…ogic (ggml-org#22456) * Refactor buffer aliasing to be part of shader lib decisions * cleanup * formatting
baramofme
pushed a commit
to baramofme/llama-cpp-turboquant
that referenced
this pull request
May 23, 2026
…ogic (ggml-org#22456) * Refactor buffer aliasing to be part of shader lib decisions * cleanup * formatting
winstonma
pushed a commit
to winstonma/llama.cpp
that referenced
this pull request
May 27, 2026
…ogic (ggml-org#22456) * Refactor buffer aliasing to be part of shader lib decisions * cleanup * formatting
fewtarius
pushed a commit
to fewtarius/CachyLLama
that referenced
this pull request
May 30, 2026
…ogic (ggml-org#22456) * Refactor buffer aliasing to be part of shader lib decisions * cleanup * formatting
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Overview
@SharmaRithik noticed that when running Granite 4.0 ssm_scan aliases several tensors, which this PR fixes by adding logic to merge those tensors into a single binding in the shader. After making that change, I realized that some of the logic for calculating aliasing could be refactored so that it is consistent across all operations and takes place in the shader library during preprocessing, so I made that change as well. I also added a test for the overlapping tensors for ssm_scan.
fyi @yomaytk
Requirements